FelixChop

os.environ["KERAS_BACKEND"] = "plaidml.keras.backend"

FelixChop

data.fillna('missing', inplace=True)
# or, for instance, if your data is numeric
data.fillna(-9999, inplace=True)

FelixChop

from tqdm import notebook
notebook.tqdm().pandas()

FelixChop

def female_proportion(dataframe):
    return (dataframe.Sex=='female').sum() / len(dataframe)

female_proportion(df)

FelixChop

df.merge(
    df.loc[
        df.Ticket.isin(
            df.Ticket.value_counts().loc[
                df.Ticket.value_counts()>1
            ].index
        )
    ].groupby('Ticket').apply(female_proportion) \
     .reset_index().rename(columns={0:'proportion_female'}),
     how='left', on='Ticket'

FelixChop

from sklearn.impute import SimpleImputer
from sklearn.preprocessing import OneHotEncoder
from sklearn.preprocessing import MinMaxScaler, StandardScaler
from sklearn_pandas import DataFrameMapper
from category_encoders import LeaveOneOutEncoder

imputer_Pclass = SimpleImputer(strategy='most_frequent', add_indicator=True)
imputer_Age = SimpleImputer(strategy='median', add_indicator=True)
imputer_SibSp = SimpleImputer(strategy='constant', fill_value=0, add_indicator=True)
imputer_Parch = SimpleImputer(strategy='constant', fill_value=0, add_indicator=True)

FelixChop

pd.DataFrame({
    'variable': variables,
    'coefficient': model.coef_[0]
}) \
    .round(decimals=2) \
    .sort_values('coefficient', ascending=False) \
    .style.bar(color=['grey', 'lightblue'], align='zero')

FelixChop

group_id, grouped_data = generator.__next__()
print(group_id) 
grouped_data

FelixChop

generator = df.groupby(['identifier']).__iter__()

FelixChop

from sklearn.preprocessing import OneHotEncoder, OrdinalEncoder
one_hot_encoder_gender = OneHotEncoder(handle_unknown='ignore')
one_hot_encoder_gender.fit(train[['Sex']])

# For Embarked column, there are some missing values. We need to first fill them then encode them.
imputer_Embarked = SimpleImputer(strategy='most_frequent', add_indicator=True)
imputer_Embarked.fit(train[['Embarked']])

transformed_Embarked = \
pd.DataFrame(imputer_Embarked.transform(train[['Embarked']]),